Mubashir Rehman
Workflows that span systems that do not talk to each other. I make them reliable, and add AI where it helps.
Rule 1: Read the number that does not add up.
Rule 2: Split diagnosis from fixing where the blast radius changes.
Rule 3: Prove the path end to end before you tune the thresholds.
Rule 8: Deterministic first, model second.
Took an AI receptionist for dental clinics from AI-scaffolded code to production, with three voice providers and per-practice tenant isolation.
The interesting part. Each practice holds its own credentials for the eligibility API, and I removed the environment-variable fallback. With a fallback, one practice's credentials could serve another. This fixes isolation at the provider boundary, not only in the data layer.
Built a client backend alone: it ingests Telegram, Discord and X, summarises with AI, searches semantically and raises alerts.
The interesting part. I wrote a JSON baseline of the existing task graph, then moved background jobs from Celery to ARQ in a planned run of commits. A queue change touches every ingestion path. The baseline gave me something to check the new workers against.
Built an agent platform over an ERP: it began as expense-claim automation and became multi-company procurement and inventory intelligence.
The interesting part. I pointed the same OCR, tool-calling and ERP layer at procurement, and moved expense and procurement onto one shared contract in a single commit. The hard parts, reading documents and acting through the ERP's API, already worked.